CN1437338A - Channel estimation method for orthogonal frequency-division multiplexing communication system - Google Patents

Abstract

A channel calculation method for communication system of orthogonal frequency division multiplexing includes 1) creating OFDM pilot frequency signal for channel calculation by system emitter with the signal including two parts of the pilot frequency insert mode and the pilot frequency value selection as well as to convert it to be the pilot frequency signal of time domain through inverse Fourier transformation, 2) transmitting the pilot frequency signal of time domain at radio multipath fading channel and 3) obtaining pilot frequency signal of frequency domain by changing the pilot frequency signal of time domain through Fourier transformation with system receiver and to carry on channel calculation for radio channel as obtaining calculation value of subchannel existed with the pilot frequency signal first by LMMSE calculation

Description

Channel estimation methods in the orthogonal FDM communication system

Technical field

The present invention relates to the channel estimation methods in the mobile communication system, relate to the channel estimation methods in the orthogonal FDM communication system particularly.

Background technology

At present, the user that develops into of mobile communication system provides the individual mobile communication terminal that can support various novel business (for example multimedia communication).Because miscellaneous service has lot of data to need transmission, so communication system requires to use higher Bit Transmission Rate.In the single-carrier system of routine, use higher bit rate, can bring difficulty for effective reception of signal because of the degree of depth frequency selective fading of intersymbol interference (ISI) and wireless channel.

A method that addresses these problems is to adopt OFDM (OFDM) technology under wireless mobile environment.In ofdm communication system, signal bandwidth less than a plurality of quadrature carriers of channel coherence bandwidth on the emission and the transmission, to overcome the signal frequency selectivity fading problem that multidiameter fading channel is brought.Can solve at interval by using protection for the ISI problem.Ofdm system has been used for digital audio broadcasting and digital video broadcasting (DVB) system in Europe at present.It also is used for Asymmetrical Digital Subscriber Line (ADSL) so that transmitting high speed data.

Common channel estimation methods is based on Adaptive Signal Processing, and wherein channel is considered to gradual.Decide according to the reception data and the channel parameter of previous moment at the channel parameter that particular moment is estimated.Under multipath fading and Quick-Change channel situation, for example in the High Data Rate OFDM mobile system,, generally adopt pilot signal to estimate to obtain the Real-time Channel response for overcoming the adverse effect of channel; Though existing channel estimation methods has high estimation accuracy, complexity is higher, and the time of estimation is longer, is difficult for realizing; So,, have necessity of improving channel estimation methods for orthogonal FDM communication system for to satisfy the needs that obtain channel estimation value under the multidiameter fading channel in the short period of time.

In ofdm system, be to finish at frequency domain based on the channel estimating and the compensation of pilot tone.Because the channel response value at data carrier is to realize by carrying out interpolation for the channel response value of pilot frequency carrier wave, so systematic function depends on the accuracy that the pilot frequency carrier wave channel response is estimated to a great extent.Have fairly simple LS to estimate based on the channel estimating of pilot tone, more complicated MMSE estimates and the time-domain and frequency-domain two dimension is united methods such as estimation.LS estimates to realize simple, but the precision of estimating is not high; MMSE estimates and other complicated methods of estimation can reach high estimation accuracy, but the algorithm complexity is difficult for realizing.Think the effective combination that reaches performance and implementation method, the difference that has proposed MMSE on the various documents is respectively simplified and approximate data.This paper proposes a kind ofly can reach certain estimated accuracy and be convenient to the channel estimation methods that the LMMSE that realizes estimates to add the filter interpolation again.

Summary of the invention

The objective of the invention is in order to overcome the contradiction of existing ofdm communication system channel estimation methods between estimated accuracy and implementation complexity, propose a kind of channel estimation methods that can satisfy in the orthogonal FDM communication system that certain estimated accuracy is convenient to again realize.

Channel estimation methods in the orthogonal FDM communication system comprises the steps:

(1) produces the used OFDM pilot signal of channel estimating by system transmitter, comprise that pilot tone inserted mode and pilot value choose two parts content; The pilot signal that produces becomes time-domain pilot signal through inverse Fourier transform;

(2) this time-domain pilot signal of emission on the wireless multipath fading channel;

(3) by system receiver time-domain pilot signal is obtained the pilot tone signal through Fourier transform; Wireless channel is carried out channel estimating, at first estimate that by the LMMSE estimation or from MMSE the simplification approximation method that obtains obtains the estimated value of pilot signal place subchannel, obtains the estimated value of whole wireless channel then with the filter interpolation method.

The inventive method the present invention is estimating OFDM communication system lower channel response effectively, and existing channel estimating will improve estimated accuracy and the contradiction that reduces between the computation complexity provides a kind of feasible method in order to solve.Fig. 4 has provided LMMSE estimation MSE least mean-square error comparison diagram under the different delayed time Rayleigh multipath channel, and as can be seen from the figure the Minimum Mean Square Error performance index of LMMSE estimation are 10 ^-2With 10 ^-3Between, effect is more satisfactory, can reach the required precision of estimation.Fig. 5 has provided the comparison that LMMSE under the Rayleigh multipath channel estimates bit error rate performance between this channel estimation methods of filtering interpolation and this channel estimation methods of LS estimation linear interpolation.Can be known that by Fig. 5 this method of LMMSE estimation filtering interpolation is estimated approach based on linear interpolation under the situation of the same error rate than LS, the signal to noise ratio improvement does not wait from 0.2 to 4dB, improves under most of situation above 1dB, and relative performance is comparatively desirable.Obtain such effect, from theory analysis, for the pilot signal of having passed through multidiameter fading channel, it is bigger that the LS method of estimation is influenced by interchannel interference ICI and Gauss's white noise AWGN, causes precision of channel estimation not high; LMMSE estimates then on the basis that LS estimates, the auto-correlation statistical property of channel and the priori of signal to noise ratio snr have been utilized, can overcome the influence of interchannel interference ICI and Gauss's white noise AWGN to a certain extent, so can be greatly improved to pilot tone place channel estimated accuracy.In addition, the filter interpolation is also good than simple linear interpolation filtering and level and smooth effect, comparatively significantly improves from 0.2 to 4dB so the effect of whole channel estimating has one.

Description of drawings

Fig. 1 is the realization block diagram of channel estimation methods of the present invention;

Fig. 2 estimates the implement device block diagram for LMMSE;

Fig. 3 is a filter interpolation implement device block diagram;

Fig. 4 is that LMMSE estimates MSE least mean-square error comparison diagram under the different delayed time Rayleigh multipath channel;

Fig. 5 is that LMMSE estimates that this channel estimation methods of filtering interpolation and LS estimate the comparison of bit error rate performance between this channel estimation methods of linear interpolation under the Rayleigh multipath channel.

Embodiment

At first produce the used OFDM pilot signal of channel estimating by transmitter, the pilot tone inserted mode adopts the Comb Pilot inserted mode, and pilot value is chosen a point on the QPSK planisphere as the constant pilot tone.The pilot signal that produces becomes time-domain pilot signal through inverse Fourier transform (IFFT).

Transmitter is transmitted into this time-domain pilot signal on the wireless multipath fading channel.

In the receiver, channel estimation methods and be achieved as follows described:

The time-domain pilot signal that receives becomes the pilot tone signal after having passed through Fourier transform (FFT), at first estimate to obtain the rough estimate evaluation of a pilot tone place subchannel, with the LMMSE linear minimum mean-squared error estimation technique rough estimate evaluation is estimated to obtain once more the thin estimated value of pilot tone place subchannel then by pilot tone signal process LS least square method; Adopt filter interpolation method to obtain frequency response on the whole channel to this thin estimated value, finish the overall process of channel estimating.

The principle of the inventive method is as follows:

The time-domain signal y (n) of the OFDM that receiver is received is by transmitting, and channel transfer function and white Gaussian noise acting in conjunction generate:

Y (n)=x (n) ^*H (n)+w (n) is passing through being expressed as on frequency domain behind the Fourier transform,

Our purpose of carrying out channel estimating of Y (k)=X (k) H (k)+I (k)+W (k) is exactly to estimate H (k) from the signal that receives.

For pilot signal be exactly

Y _P(k)=X _P(k) H _P(k)+I _P(k)+W _P(k) X _p(k) be known pilot signal value, Y _p(k) signal value for receiving, H _p(K) be channel frequency response value on the subcarrier, W _p(k) be white Gaussian noise AWGN on the subcarrier, I _p(k) disturb ICI between the subchannel that is.

Channel estimating based on pilot tone can be divided into for two steps, at first was to estimate H _p(k) value is then to H _p(k) carry out interpolation, recover the value of H (k).Realization block diagram of the present invention at first is to carry out LMMSE at 101 pairs of pilot tone place subchannels to estimate as shown in Figure 1, is then to obtain the estimated value of whole channel 102 by the filter interpolation.

Its specific implementation step and method are described below:

After time-domain signal transfers the signal of frequency domain to by Fourier transform,

(1) at first obtains the rough estimate evaluation of channel impulse response at the receiver of ofdm communication system by LS (Least Squares) least square method ${\hat{H}}_{p,\mathrm{ls}}=\[H_{p,\mathrm{ls}}\left(0\right),H_{p,\mathrm{ls}}\left(1\right),\…,H_{p,\mathrm{ls}}(N_p-1){\]}^T$ $=X_p^{-1}{Y}_p$ $=\[\frac{Y_p\left(0\right)}{X_p\left(0\right)},\frac{Y_p\left(1\right)}{X_p\left(1\right)},\·\·\·,\frac{Y_p(N_p-1)}{X_p\left(N_{p-1}\right)}\]$

X wherein _p(0), X _p(1) ..., X _p(N _p-1) pilot tone for sending, N _pNumber for the pilot tone that sends. Be the correspondence position of pilot signal at subcarrier.

(2) on the basis of LS rough estimate, estimate to obtain the thin estimated value of channel impulse response then by LMMSE (Linear MMSE)

Average by doing to transmitting, use static pilot value X _P(m)=and c, m=0,1 ..., N _P-1 Linear MMSE method of estimation that obtains to simplify is, $\widehat{H_P}=R_{H_p{H}_p}{(R_{H_p{H}_p}+\frac{\mathrm{\&beta;}}{\mathrm{SNR}}I)}^{-1}{\hat{H}}_{p,\mathrm{ls}}$ Wherein, $R_{H_p{H}_p}=E\left\{H_p{H}_p^H\right\},$ Be the autocorrelation matrix of channel response, can send training sequence by transmitter and obtain or obtain by prior statistical measurement by the receiver actual measurement; $\mathrm{SNR}=E|X_P\left(k\right){|}^2/{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="A0311874600111.png,A0311874600112.png"\; state="\{\{state\}\}">n</figure-callout>}}^2$ Be the signal to noise ratio of pilot carrier signal, can measure in real time at receiver; β=E|X _P(k) | ²E|1/X _P(k) | ²For depending on the constant of modulated signal constellation, for QPSK modulation, β=1.Its specific implementation and device block diagram are as shown in Figure 2.

(3) the thin estimated value of the channel impulse response that at last LMMSE is obtained is carried out the impulse responses that filter interpolation (FilterInterpolation) obtains whole channel

The realization of filter interpolation and device block diagram are as shown in Figure 3, earlier two adjacent pilot tone estimated values are carried out the interpolation that integral multiple is I, promptly between two adjacent pilot tones, insert I-1 0 value point, carry out low-pass filtering then, draw the channel response value of data carrier between adjacent two pilot tones like this with regard to interpolation, thereby obtained the impulse responses of whole channel.

From theory analysis, the MMSE method of estimation ${\hat{H}}_{p,\mathrm{MMSE}}=R_{H_p{H}_{p,\mathrm{ls}}}{R}_{H_p{H}_{p,\mathrm{ls}}}^{-1}{\hat{H}}_{p,\mathrm{ls}}$ $=R_{H_p{H}_p}{(R_{H_p{H}_p}+{\mathrm{\&sigma;}}_n^2{\left(X_p{X}_p^H\right)}^{-1})}^{-1}{\hat{H}}_{p,\mathrm{ls}}$ Need to carry out twice matrix inversion operation and matrix multiplication operation altogether, amount of calculation and handle comparatively complicated; The LMMSE method of estimation then only needs matrix inversion operation and matrix multiplication operation, obtain and the basis of the estimation effect that MMSE is more approaching on, the processing complexity reduces greatly.

Below, the present invention is further detailed explanation with example.

At first produce the used OFDM pilot signal of channel estimating by transmitter, pilot tone inserted mode employing pilot tone and data are 1: 3 Comb Pilot inserted mode, and pilot value is chosen a point on the QPSK planisphere, and present embodiment is got 1+j as pilot signal.The pilot signal that produces becomes time-domain pilot signal through inverse Fourier transform (IFFT).

Transmitter is transmitted into this time-domain pilot signal on the wireless multipath fading channel.

In the receiver, channel estimation methods and realizing as shown in Figure 1:

Receiver carries out becoming the pilot tone signal behind the Fourier transform (FFT) to time-domain pilot signal, extracts pilot signal by the pilot tone insertion position of knowing in advance; This pilot signal inputs to the 101LMMSE estimation module, obtains the thin estimated value of channel and gives 102 filter interpolation, exports the impulse responses of whole channel.

Corresponding to the 101LMMSE estimation module implementation procedure of Fig. 1 as shown in Figure 2, be divided into two-way by statistics priori channel impulse response value 201, one the tunnel multiplies each other at 203 multipliers with untreated another road at 202 matrix operation device transposition and after asking conjugation obtains the channel autocorrelation value, asks through 204 weighted accumulation devices to obtain the channel autocorrelation matrix after average

Multiply by the gain of receiver measurement gained with the unit matrix of 205 storages 206 Amass in 207 adder additions, after the addition and obtain its inverse matrix at 208 matrix operation devices; 209 receive pilot signal obtains LS channel rough estimate evaluation 211 divided by reference pilot value 210 (being 1+j in this example); 204 channel autocorrelation matrix

Multiply each other at 212 multipliers with the 208 LS channel rough estimate evaluations that obtain result and 211 and to obtain the thin estimated value of LMMSE channel and by 213 these channel estimation values of output.

Realize as shown in Figure 3 corresponding to Fig. 1 median filter 102 interpolation methods, pilot sub-carrier channel estimation value for 101 outputs among Fig. 1, in the null value interpolater 301 of Fig. 3, at first carry out 4 times up-sampling, promptly between per two pilot sub-carrier channel estimation values, insert three zero, obtain the impulse responses of whole channel after burst process low pass filter (being LPF) the 302 filtering interpolation behind the up-sampling, LPF is the FIR low pass filter on one 51 rank in the present embodiment.

In the inventive method, the pilot tone inserted mode can also be the interleaved plum blossom shape pilot tone inserted mode of pilot frequency locations of front and back time slot except the Comb Pilot of fixed-site.Simulation results shows the difference according to wireless multipath channel signal to noise ratio, the ratio of pilot tone and data-signal (1: except being 1: 3, can also be 1: 4 and 1: 5 in the channel signal to noise ratio than (for example greater than 8dB) under the condition with higher n).

The major part LMMSE of this paper channel estimation methods estimates it is the simplification approximation method that MMSE estimates; The present invention can obtain the effect of akin channel estimating for other linearities (for example polynomial approximation algorithm of MMSE) or nonlinear MMSE method of estimation.

But the interpolation filter that the present invention uses is the higher linear phase of an exponent number FIR filter preferably; Better when channel condition, the receiver phase deviation estimated of compensate for channel or when requiring implementation complexity lower easily can be adopted linear phase weaker but iir filter or other filters that exponent number is lower.

Claims (6)

1. the channel estimation methods in the orthogonal FDM communication system comprises the steps:

(1) produces the used OFDM pilot signal of channel estimating by system transmitter, comprise that pilot tone inserted mode and pilot value choose two parts content; The pilot signal that produces becomes time-domain pilot signal through inverse Fourier transform;

(2) this time-domain pilot signal of emission on the wireless multipath fading channel;

(3) by system receiver time-domain pilot signal is obtained the pilot tone signal through Fourier transform; Wireless channel is carried out channel estimating, at first estimate that by the LMMSE estimation or from MMSE the simplification approximation method that obtains obtains the estimated value of pilot signal place subchannel, obtains the estimated value of whole wireless channel then with the filter interpolation method.

2. method according to claim 1 is characterized in that: the pilot tone inserted mode in the step (1) adopts the Comb Pilot inserted mode.

3. method according to claim 2 is characterized in that: in the described Comb Pilot inserted mode, pilot tone is the 1 intensive pilot tone inserted mode than n with the ratio of data-signal, n=3,4,5.

4. according to claim 1,2 or 3 described methods, it is characterized in that: the filter interpolation in the described step (3) comprises null value interpolation and low-pass filtering successively.

5, method according to claim 4 is characterized in that: described low-pass filtering adopts FIR filter or iir filter.

6. method according to claim 4 is characterized in that: described null value interpolation is that adjacent pilot frequencies signal place subchannel estimated value is carried out the doubly interior zero insertion of n, and wherein n is the multiple of data-signal for pilot signal.

Images